Response by the Society of American Archivists to NARA
Advanced Notice of Proposed Rulemaking
Prepared for SAA by its Electronic Records Section and its Government Records Section,
and Endorsed by SAA Council
4 January 2002
Summary
Retaining the content, structure and context of records is a necessary component of
responsible recordkeeping. When transferring records from one technical environment to
another, one can potentially lose essential components of that content, context and
structure. Appropriate management, retention and disposition of electronic records
requires government agencies -- with guidance and legal authority from the National
Archives and Records Administration -- to address the necessary elements and properties
of the particular forms and formats of those records. We believe that Public Citizen's call
for the retention of "the entire content, structure and context of the electronic original" is
neither realistic nor helpful as a goal of good recordkeeping. Instead, agency policies,
procedures and documentation -- including NARA-approved retention schedules,
published disposition manuals, special transfer requirements, transfer forms, and other
documentation -- should indicate the specific qualities of digital objects that must be
retained in order to ensure their ongoing value as records.
While we support most of the requirements that NARA provides in 36 CFR 1234.22, we
believe that they should be applied to all records produced or received with electronic
information systems. We do not see "text documents" as a useful category for purposes
of this regulation, since we do not believe that it is possible to identify any particular set
of elements or properties that must be retained for all records that fall within its
definition.
We also emphasize that retaining records in electronic formats will often strongly support
the business needs of government, including provisions for public access. As Public
Citizen argues, retention of records in electronic formats will often greatly promote both
the convenience and quality of public access to those records. Recent case law provides a
number of examples in which parties have been required to provide the electronic version
of records as part of the evidence discovery process. While this does not imply that all
records must be retained in electronic formats for their entire life, it does provide
compelling support for the view that electronic source records can often contain
important elements of evidence that may be lost in printed copies.
NARA's mission is ensuring "ready access to essential evidence." This is a primary
objective of the archival profession as a whole. It is also a mission in which the federal
government, citizens, and other potential researchers all have an interest. EFOIA
represents an important set of business needs related to such access. Recent legislation
2
such as the Government Paperwork Elimination Act and Electronic Signatures in Global
and National Commerce Act has also introduced a number of requirements to manage
and provide access to government records in electronic formats. Of course, agencies
must continually strike a balance among numerous business needs. This balance is
recognized in the FOIA legislation, which states that each "agency shall make reasonable
efforts to maintain its records in forms and formats" that can be both searched and
reproduced electronically. We encourage NARA to continue expanding its efforts to
promote the management and preservation of electronic records within the federal
government.
Introduction
In response to the request for comments on proposed rulemaking published in the Federal
Register on 10 October 2001 and identified as 3095-AB05, the Society of American
Archivists (SAA) submits the following document.1 The Society of American Archivists
(SAA) is the oldest and largest association of professional archivists in North America.
Representing more than 3,000 individuals and 400 institutions, the SAA is the
authoritative voice in the United States on issues that affect the identification,
preservation, and use of historical records.
Reconciling existing laws, regulations, and principles with the new demands of records
created and received by new information technologies is a great challenge facing the
archival profession. The SAA commends NARA on its recent and continuing efforts to
provide further guidance on the management and preservation of electronic records.
NARA’s response to the Public Citizen petition and its solicitation of comments on that
petition is indicative of NARA’s willingness to engage with the larger archival
community in discussion of these important issues.
General Comments on Public Citizen's Petition
Public Citizen's petition1 rests on the claim that it is important to preserve records in their
original form. The professional literature of archival science supports the general
principle that at times it is appropriate to maintain records in their original form. One of
the most compelling reasons is that originals have qualities that may not be reflected in
1
Two constituent units of the SAA, the Electronic Records Section and the Government Records Section,
are responsible for most of the comments included here. The following individuals from the two sections
contributed texts and comments during the process of drafting this document: Paul Bergeron, City Clerk,
Nashua, New Hampshire; Jelain Chubb, Missouri State Archives; Christopher Frey, University of
Michigan; Geof Huth, New York State Archives; Craig Kelso, Missouri State Archives; Joe Laframboise,
Kansas State Historical Society; Cal Lee, University of Michigan; Scott Leonard, Kansas State Historical
Society; Pat Michaelis, Kansas State Historical Society; Marry-Ellyn Strauser, Missouri State Archives;
Matt Veatch, Kansas State Historical Society; and David Wallace, University of Michigan. The Electronic
Records Section and the Government Records Section both include members who are employees of the
National Archives and Records Administration. Those members have abstained from the process of
drafting this document.
3
facsimiles. These qualities can potentially serve a variety of business, legal, audit,
operational and research needs. The need to retain originals, however, must be weighed
against institutional priorities, available resources, and the limits of technical feasibility.
It is thus important to provide very specific answers to the question, "What qualities of an
original are useful or necessary to retain in their original form?"2
For physical records, archivists have stipulated that records with "intrinsic value" are
those records worth preserving in their original form. Intrinsic value was defined in a
1974 SAA glossary as embodying the "qualities and characteristics of permanently
valuable records that make the records in their original physical form the only archivally
acceptable form of the records."3 In 1982, the U.S. National Archives (then NARS)
issued a paper on intrinsic value, explaining that these "qualities or characteristics relate
to the physical nature of the records, their prospective uses, and the information they
contain."4
Although the "essential intangibility"5 of electronic records makes it inappropriate to
focus on their physical characteristics in the same way one would for paper documents,
the question of which properties to preserve becomes even more important. As indicated
by current NARA policies,6 the long-term preservation of electronic records will
generally require copying (also known as refreshing) and reformatting onto new physical
storage media, 7 but it is important that one "not alter the character of the recorded
information" in the process.8 This requires one to preserve qualities at a higher level of
abstraction than the physical medium. Though it is debatable whether or how the concept
of an "original" is still relevant in this context,9 there must still be some way of indicating
the characteristics of an electronic source record that should be retained in any
recordkeeping copies. This is supported by the revised definition of intrinsic value in the
SAA Glossary from 1992, which refers to the "inherent worth of a document" rather than
its original physical form.10
The identification of content, structure and context as essential characteristics of records
is thus extremely important. If properly designed and managed, a recordkeeping system
can retain these characteristics of records even though underlying hardware and software
may change over time. As Public Citizen has pointed out, the need for a recordkeeping
system to retain records content, structure and context for their required retention period
was emphasized in both of the legal cases related to GRS 20. "In other words, as counsel
for the Archivist put it at oral argument, if the information is part of a record under the
RDA [Records Disposal Act], see § 3301, then it must be preserved."11 This still leaves
the difficult task of determining what specific pieces of information fit this description.
We believe that Public Citizen's call for the retention of "the entire content, structure and
context of the electronic original" [emphasis added] is not realistic. Instead, agency
policies, procedures and documentation -- including NARA-approved retention
schedules, published disposition manuals, "special transfer requirements agreed upon by
NARA and the agency [to] be included in the disposition instructions,"12 transfer forms,
and other "documentation adequate to identify, service and interpret electronic records
that have been designated for preservation by NARA"13 -- should indicate the specific
4
qualities of digital objects that must be retained in order to ensure their ongoing value as
records.
A related concept, introduced by the CEDARS project on digital preservation, is that of
"significant properties." According to researchers on that project, "Whoever takes the
decision that a particular digital object should be preserved will have to decide what
properties are to be regarded as significant." Within the terminology of the Reference
Model for an Open Archival Information System (OAIS), the "representation
information" related to a digital object "must allow for the recreation of the significant
properties of the original digital object."14 Other recent work on digital preservation has
reiterated and reinforced the important role of significant properties.15
For the management and preservation of federal records, this would imply the need to
clarify the properties of "form or character" of electronic source records that either serve
a sufficient business need to warrant preservation by the creating agency or have
"sufficient administrative, legal, research, or other value to warrant their further
preservation"16 through transfer to the National Archives. It is worth reiterating that for
electronic records, the properties of this form should be defined independent of
underlying physical medium, whenever possible. This can allow for the creation of
recordkeeping copies of an electronic record on various media, including new electronic
platforms, and -- often but not always -- paper, microfilm or other analog media. Without
the specification, capture, management, and verification of such properties, however,
there is no guarantee that transfer across media will retain its value as a record of
government business.
Public Citizen’s Proposal 1
“The regulations should make explicit that recordkeeping systems that preserve
electronic text documents must preserve the entire content, structure and context of
the electronic original, a requirement that the Archivist’s attorneys have stated is
already part of GRS 20, although the text of GRS 20 contains no such language.”
1A1. Is NARA’s definition of electronic information system still adequate?
We would suggest the following wording:
Electronic information system. A computer-based system that supports a
set of one or more of the following functions: acquisition, creation,
storage, processing, management and access to information.
This system will often include application programs, business applications, system
software (including supporting databases and servers), desktop operating systems,
network operating systems and servers. In order to clarify further the role of an
electronic information system, we offer the following revised definitions:
5
Electronic recordkeeping system. An electronic information system that
supports the collection, organization and categorization of records to
facilitate their preservation, retrieval, use and disposition.
This definition indicates that electronic recordkeeping systems are the subset of
electronic information systems that provide recordkeeping functions. The system itself is
electronic, but all of the records to which it is applied need not be electronic. If NARA
would instead like the term "electronic recordkeeping system" to mean that subset of
recordkeeping systems that manage electronic records, then we would suggest a wording
that more explicitly states this relationship to specifically electronic records. The current
36 CFR 1220.14 definition, however, is ambiguous. A reader could also potentially
interpret the wording of "system in which records..." to mean that an electronic
recordkeeping system must be the electronic information system that physically stores
electronic records. We do not believe that such a definition would be useful, since
records management applications and other software that supports recordkeeping need
not provide this function directly. The important point to emphasize about an electronic
recordkeeping system is that provides important functionality that not all electronic
information systems provide, regardless of what combination of hardware and software
may support that functionality at any given point in time.
Recordkeeping system. A system in which records are collected,
organized, and categorized to facilitate their preservation, retrieval, use,
and disposition. The system can contain manual and automated
components.
This wording eliminates the "manual or automated" distinction of the existing 36 CFR
1220.14 definition, which is potentially misleading. As indicated by the Electronic Work
Group Report, a system consists of "people, machines, and methods organized to
accomplish a set of specific functions." Given these necessary components, a
recordkeeping system thus will never be fully automated. Conversely, in a contemporary
government office environment, even a recordkeeping system that manages records that
all reside on paper media will still include numerous automated (or at least electronic)
components -- indexes, forms, directories, etc. -- in order to support the collection,
organization and categorization of those records.
Should it explicitly include (or exclude) any types of office applications or other
types of software such as the network operating system?
An electronic information system will tend to include one or more applications, and it
will always rely on some components of one or more operating systems. Though a
longer narrative explanation of electronic information systems could provide examples of
such components, it would be both unnecessary and inappropriate to delineate them
within the definition itself. The specific functional roles played by the layers of
supporting technology will vary across technical environments at any given time, and
they will vary even more dramatically across time. In order to implement the
functionality of an electronic information system one needs a large number of low-level
6
activities -- processing of machine instructions, process management, remote procedure
calls, load balancing, memory management, file management, directory management,
input-output, storage allocation, data typing, compression, encryption, error correction,
interfaces in between various software processes, access control, data caching, data
locking, journaling, processing of user commands, desktop and windowing, etc. -- which
can be carried out by any number of software or hardware components. It would be a
mistake for the definition to attempt a coherent, exhaustive delineation of such
components.
Is the definition of “text documents” sufficiently broad enough to cover documents
produced by products other than word processing software, e.g. PowerPoint
presentations or desktop publishing files?
We believe that "text document" is not a useful category for purposes of 36 CFR 1234.
One can target specific structural elements for retention in the case of both relational
database tables (field names, field lengths, foreign keys, primary keys, etc.) and email
messages (required MIME header fields, multipart message boundaries, etc.). This is
because each has a fairly "strictly prescribed form and format."17, 18 The same cannot be
said for those digital objects designated under the current definition of "text documents."
In fact, the definition itself indicates that they have a "loosely prescribed form and
format." Allowing disposition of electronic source records based on this definition could
lead to the ironic and problematic situation in which the electronic records whose
properties are least formally understood (and thus most difficult to identify or verify
during transformation) can be destroyed after being transferred to a new medium.
We would suggest instead that NARA change the name of 1234.22 to "Creation and use
of electronic records" and move it to the beginning of Subpart C, since it would appear
that its provisions are appropriate for all types of electronic records. The general
description of electronic records in 1234.1 indicates that they "include numeric, graphic,
and text information," all of which are possible components of files created through
commercially available office applications. Given the tight integration of many suites of
office applications, anyone applying the definition of text documents to all such files
would seem unable to exclude interactive content, macros, images, tables, spreadsheets,
animation, hyperlinks and various multimedia components without knowing more about
the specific "form and character" of the specific category of records.
If disposition specific to text documents is provided, we would encourage clarification of
the distinction between this definition of text documents and that of "textual documents"
in 36 CFR 1228.270(d)(2). The latter provides for only ASCII files (which may contain
SGML markup), obviously excluding a large number of current office documents without
further transformation. In that same light, we would also encourage clarification of the
distinction between "data files" as defined in 1234 and 1228. In 36 CFR 1228.270(e)(3),
for example, "textual documents with SGML tags shall include a table for interpreting the
tags, when appropriate." Since software-dependent database files are often likely to be
translated into SGML (or, more likely, XML) for purposes of long-term preservation
7
(either within an agency or when transferred to the National Archives), further
explanation of the role of electronic recordkeeping copies may be helpful.
Should the definition of “text documents” be amended to include presentation
and other specific files?
We recommend avoiding the delineation of specific examples within definitions of terms,
since the list can often be inappropriately perceived as either exclusive or exhaustive.
1A2. If we determine that the section should be amended to reflect Public
Citizen’s proposed requirements, would coverage of the section be clearer if the
term “electronic information systems” is replaced in §1234.22 by a delineation of
specific applications that may produce original electronic text documents such as
office suite application packages (e.g., Office 2000, Lotus Notes), or word
processing or other office automation applications not integrated with the agency
email or office suite?
Once again, we recommend avoiding the delineation of specific examples within
definitions of terms, since the list can often be inappropriately perceived as either
exclusive or exhaustive.
1B1. Are the definitions of “content,” “structure,” and “context” contained in the
NARA GPEA guidance adequate for all types of records?
Since all electronic records have content, structure and context, this definition provides
further support for moving the current 1234.22 to the beginning of Subsection G and
applying it to all records produced or received with electronic information systems, rather
than simply text documents.
NARA defines “content” as “The information that a document is meant to convey.
Words, phrases, numbers, or symbols comprising the actual text of the record that were
produced by the record creator.” We would suggest that the second sentence narrows the
definition unnecessarily. By explicitly including the terms “words, phrases, number, or
symbols,” it implicitly excludes numerous other types of content (e.g. pictorial, aural).
As with other items above, we would not recommend revising to attempt a fully inclusive
list of possible content types. Rather, we would advise using only the first part of this
definition, which is identical to the definition found in the current SAA Glossary.19
We find the definition of structure to be adequate.
The first part of the definition of context is consistent with our recommendation of a
broad conception of a recordkeeping system: “The organizational, functional, and
operational circumstances in which documents are created and/or received and used." It
also emphasizes why we do not support Public Citizen's call for the preservation of the
"entire content, structure and context" of records over time. In principle, the context of
any given record is infinitely complex. Effective recordkeeping systems are necessary to
8
ensure the capture, management and preservation of an important portion of record
context. The alternative is independent data and information elements that cannot
adequately stand as evidence of government activities.
The second part of the definition of context emphasizes those components for which
electronic systems are often very well suited: "The placement of records within a larger
records classification system providing cross-references to other related records.” Though
both parts are important, we would suggest possibly combining them to form a much
more compact definition:
Context: The organizational, filing and classification systems in which
documents are created, received and used.
This definition also eliminates “operational,” which may be redundant. Though more
succinct, we do recognize that this definition may be a bit less transparent and thus may
not necessarily be preferable to the existing definition. We also believe that an
explanation of context should include technical contextual elements (metadata, software
documentation, etc.) as well as the business context. The current definition could be
interpreted to include these elements, so it is not completely clear whether they should be
added to the definition itself, though we would recommend that any supporting
documentation contain some mention of them.
We would like to reemphasize that the retention of all context of electronic source
records is not a desirable or realistic goal. Not all contextual information is of equal
value. The organization of email within the inbox of its recipient is far less important
than contextual information tied specifically to the filing and use of the message.
Do you agree with NARA’s understanding of the terms “content,” “structure,” and
“context” as they apply to text documents in the Federal Government? If not,
what is your understanding of these terms?
Apart from a few minor suggestions provided above, we find the general definitions of
content, structure and context to be adequate. We do take issue with the category of text
documents. It is both appropriate and desirable for NARA to provide technical guidance
on important properties of particular formats that are consistently applied within those
formats (e.g. attachments to electronic mail messages), and it is further desirable to
identify forms20 of electronic documents that constitute specific records series with their
own associated properties. Form of material (e.g. time sheet, birth certificate) will often
indicate the appropriate retention period for a given record -- as codified through a
NARA-approved retention schedule series or published agency disposition manual -while the electronic format can often indicate specific properties that hold for all files
created in that format (e.g. To and From fields in the MIME header of all electronic mail
messages) and others that only hold on occasion (e.g. document property fields that are
used for word processing documents on some occasions by not on others). The
combination of the two can indicate not only what records should be retained, but also
what properties should be retained in the creation of an electronic recordkeeping copy
9
from an electronic source record or the creation of an analog recordkeeping copy from an
electronic source record. Primarily for reasons already stated, however, we do not
believe that guidance on content, context and structure should be applied to text
documents as a general category.
Do these concepts need to be defined in NARA regulations?
Yes. The concepts of content, structure and context are essential to understanding and
managing records, especially electronic records. Their definitions should be included in
1234.2, and they also should be explicitly referenced in the section being addressed by
NARA's ANPRM (current 1234.22) as recommended by Public Citizen.
1B2. What information about the content, structure, and context must be
maintained as part of the record for the agency to conduct its business and for
accountability purposes?
We believe that it is impossible to provide a specific answer to this question that would
apply to all records. We can provide a few general base-line suggestions.
Content: For any essential data element of an electronic record21 agencies should take
appropriate measures to maintain the integrity of the bit/byte stream of that element if it
is preserved electronically and ensure that the full contents of that data element get
represented physically in some way if the recordkeeping copy is to be on an analog
medium.
Structure: Agencies should preserve all the structure of a document that is necessary to
make the content sufficiently unambiguous and meaningful as evidence.
Context: At the very least, agencies must create and preserve enough contextual
documentation of a record to associate it with a given business function, activity and/or
transaction.
Can we define the minimum metadata needed for text documents to provide
adequate documentation, as we do for email messages (see 36 CFR
1234.24(a)(1)(a)(3))?
One could provide the minimum metadata to provide for adequate documentation of all
records, but it is not possible to identify any specific set of additional metadata that
should apply to text documents as a category.
Are the minimum metadata different for permanent and temporary records?
There will be additional metadata required for records that are preserved for long periods
of time. This will include technical and administrative metadata, as well as
documentation to help a user far into the future make better sense of the record. Most of
this metadata will be added over time, however, and not at the point of creation. It is
10
likely that one will want to add some additional elements to a record that needs to be
retained for longer than the expected life of the current hardware and software
environment. This distinction, however, is purely based on technological requirements,
rather than any division between temporary and permanent records based on appraised
value.
Do specific types of text documents require different minimum metadata?
If type is taken to mean a different form or format of records, then it would be possible to
specify different elements for different types.
What relationship do you see between “content, structure, and context” and
metadata requirements?
The line between data and metadata is not absolute. The important thing is to identify
necessary elements and properties. Depending on the implementation of an agency, these
may be retained as part of digital objects themselves or as external elements related to
those objects.
Specifically addressing the Public Citizen proposed CFR wording, does
compliance with the metadata and other requirements in this proposed §
1234.22(a)(5) meet the requirements for content, structure and context in its
proposed § 1234.22(a)(1)?
These requirements are meeting different sets of objectives, and thus should both be
included. Content, structure and context are necessary to preserve an authentic record.
Whether their constituent elements are considered data or metadata will often come down
to agency implementation. The requirements of 1234.22(a)(5) may be met, at least in
part, by many of the same elements retained for purposes of 1234.22(a)(1), but (a)(5) will
tend to require additional descriptive and administrative elements specific to its purposes.
Retrieval of an electronic record, for example, will generally require things such as an
identifier that is unique to the system and some sort of browsing or querying capability.
The data elements necessary to perform such functions will most likely not reside within
the electronic source record, but must instead be added as part of the recordkeeping
system. Metadata for enabling retrieval, protection and disposition is neither necessary
nor sufficient for the preserving the content, structure and context of the record itself.
1B3. Hidden Information: NARA’s view is that hidden information (such as
comments) in text records must be preserved as part of the record when the
author intends to share the information with others, e.g., notes added to explain
or comment on a draft report.
According to 36 CFR 1222.34, Identifying Federal records:
11
(c) Working files and similar materials. Working files, such as preliminary
drafts and rough notes, and other similar materials shall be maintained for
purposes of adequate and proper documentation if:
(1) They were circulated or made available to employees, other
than the creator, for official purposes such as approval, comment,
action, recommendation, follow-up, or to communicate with
agency staff about agency business; and
(2) They contain unique information, such as substantive
annotations or comments included therein, that adds to a proper
understanding of the agency's formulation and execution of basic
policies, decisions, actions, or responsibilities.
For documents that should be retained based on the conditions stated above, retaining
annotations and comments would be necessary. It would seem that the second condition
is provided specifically to indicate cases beyond those in which the author specifically
"intends to share" the contents of the annotations or comments with others. The next
paragraph of that same section would seem to support the idea that an electronic record
that is used to carry out a given activity can have record status even if a paper version
also exists.
(d) Record status of copies. The determination as to whether a particular
document is a record does not depend upon whether it contains unique
information. Multiple copies of the same document and documents
containing duplicative information, including messages created or
received on electronic mail systems, may each have record status
depending on how they are used to transact agency business.
Though NARA goes on to explain in (f)(2) that extra copies of documents that are
preserved "solely for convenience of reference" should be considered nonrecord material,
it is by no means obvious that all (or even the majority of) use of electronic documents
for which a paper version exists actually fall into this "convenience copy" category.
According to NARA's Introduction to its General Records Schedules, agencies are
instructed, "When it is difficult to decide whether files are record or nonrecord materials,
the records officer should treat them as records."22
We support the Electronic Records Work Group's use of the term "electronic source
record," since the guidance cited above and the definition of record at 44 USC 3301
would seem to imply that a large proportion of electronic documents created by federal
agencies are records. This does not mean that all such records should have long retention
periods, nor does it mean that they all contain elements (such as comments or
annotations) that would not be reflected in a paper printout. We would argue, however,
that it does imply a need to address such issues at a much finer level of granularity than
all "text documents." It is incumbent upon agencies, with guidance from NARA, to take
measures to minimize the risk of destroying essential evidence through the deletion of
electronic source records.
12
Is it essential or even misleading to require it when the document is
viewed/printed from a system that does not indicate that there is hidden text?
Data stored on electronic media requires hardware and software to be discovered,
accessed, read and understood. It is thus all "hidden" from us in a way that ink on paper
is not.23 A specific view of an electronic record using a specific application, given a
specific set of user preferences on a specific computer platform is thus often not the
appropriate standard for determining what to count as the "contents" of that record, nor,
often, is printing that same record out onto paper using that same application, set of
preferences and computer platform. A significant proportion of electronic formats
(including those created by most common office suite applications) support multiple
"views" of the same document. For some formats, one may be able to determine a
general canonical view that can be assumed to reflect all significant properties of the
"record version" of the document, but for a large number of others, such a universal
determination would be both arbitrary and likely to neglect essential evidence of
government business.
What types of documents besides word processing have hidden comments/text
capability, e.g., spreadsheets with formulas?
Most electronic document formats include numerous forms of "hidden" content, as it is
understood in the question above. Spreadsheets are no exception.24 They can include
formulae, macros, notes of various kinds, embedded complex objects, links to external
data sources and often reflect dependencies between individual sheets within a given file.
They also support a large number of formatting options (e.g. width and wrapping of cells,
hiding of values, highlighting, generation of charts) and multiple views of any given
sheet. The rendering of the spreadsheet on screen as well as the content and appearance
of a printouts will vary considerably, based on factors such as operating system,
spreadsheet application and version, user preferences, fonts supported, driver software
and rendering hardware.
Document summaries: What elements of document summary information are
commonly available from all major word processing applications?
Details vary from one application to another, but some examples from Microsoft Word
97 include:
• Title
• Subject
• Author
• Manager
• Company
• Category
• Keywords
• Comments
• Hyperlink Base
13
•
•
•
•
•
•
•
•
Date and Time Created
Date and Time Modified
Date and Time Accessed
Date and Time Printed
Last Saved By
Revision Number
Total Editing Time
Various document statistics and possible custom properties
It is also worth noting that many commercial email applications also store numerous
elements that users will generally not see unless they make an additional effort to look for
them. Many of these are created and retained automatically, and can often have
important recordkeeping value. John Jessen, Managing Director of Evidence Associates,
points out some of the noteworthy elements from a standard Microsoft Outlook email
message:25
Attachment
BCC
Billing Information
Categories
CC
Changed By
Contacts
Conversation
Created
Defer Until
Do Not AutoArchive
Download State
Due By
Expires
Flag Status
Follow Up Flag
From
Have Replies Sent To
Icon
Importance
In Folder
Internet Account
Junk E-mail Type
Message
Message Class
Modified
Outlook Internal Version
Outlook Version
Read
Receipt Requested
Received
Relevance
Remote Status
Retrieval Time
Sent
Size
Subject
To
Tracking Status
Other summary information can be stored in locations such as templates, the file system
of the operating system, directory services of a network operating system and supporting
system files (e.g. the Windows Registry). We believe that a survey of the most popular
office suite applications would probably reveal a number of properties that they have in
common.
What other office applications that produce text documents have a similar
feature?
Office suites tend to have many properties that can be applied to all file formats, with a
few that are format-specific.
Is the document summary feature used in your agency and, if so, how widely?
Does any agency require staff to complete the document summary routinely?
14
Since we are not representing a specific agency, we cannot answer this question as it is
stated. Many anecdotal accounts would seem to indicate that conscious, extensive use of
summary features on an agency-wide basis is rare. Basic principles of technological
adoption would also suggest that users will not exert extra effort in using the features
unless they have an incentive to do so. The only way to truly determine the extent of use
would be collect empirical data at more than an anecdotal level.
It is important to note, however, that many features can be retained automatically.
Electronic document management systems (used in some federal agencies) also often
make use these native features of office suite applications.
Is a default normally used?
Without any incentive to do otherwise, it is likely that most users do not take any action
to change document property values from the default. As noted previously, some are
assigned automatically. Given the presence of an EDMS and/or records management
application, users have considerably more incentive to use such features.
How does the agency use the information if they retain the document in a nonelectronic recordkeeping system?
We would speculate that this fairly uncommon, though some of these properties can
appear automatically as part of a print out, either as part of a cover sheet or along the
margins of a document.
1C. If we determine that § 1234.22 should be amended to reflect Public Citizen’s
proposed requirements, should we retain the current paragraph (a)(3) for
electronic recordkeeping systems only?
Yes. Since standard interchange formats can be important to numerous forms of
electronic records, this provides yet another reason to move the existing 1234.22 to the
beginning of Subpart G.
1D. Do you see any other issues that should be considered as we evaluate the
Public Citizen Proposal 1?
We share Public Citizen's concern that agencies often do not comply with their stated
policies to "print and delete" records. We support the rationale behind Public Citizen's
proposed wording for 1234.22(a) to include "recordkeeping systems that maintain the
official copy of text documents produced on electronic information systems" rather than
only "electronic recordkeeping systems that maintain the official file copy." Given our
concerns about "text documents" as a category, however, we would recommend wording
such as the following: "recordkeeping systems that maintain the official copy of records
that were produced or received with electronic information systems."
15
NARA's current wording could be interpreted to allow agencies to output records to
hardcopy (paper, microform, etc.) without adhering to any of the other requirements of §
1234.22. Public Citizen’s proposed revision explicitly requires agencies to provide for
access, security, disposition and metadata standards for any records created
electronically. We would also encourage NARA to include references to the following in
its guidance on the disposition of electronic sources records: provisions for policy
implementation auditing and periodic quality control verification of recordkeeping copy
production.
Proposal 3
“Public Citizen states in its petition that records in electronic form have unique
advantages, including wider and easier distribution, searching and indexing the
records, and storage. Public Citizen further states that 'electronic records carry
advantages for research, even if the records have not been maintained in a system
that satisfies all of the attributes of an ideal electronic recordkeeping system.'
Public Citizen argues that it is important to address the disposition of both text
documents and data files whenever new information systems are developed.”
3A1. Does (and should) electronic information “system” as used in this proposed
paragraph include word processing applications?
Yes.
If so, does the word processing application technically “store” the text documents
produced with the software?
No.
3A2. Should we distinguish systems that only produce or use electronic records
from those that store them?
We do not believe that this is necessary. An electronic information system will consist of
numerous components. For purposes of this document, it is not relevant whether
production and storage are performed by the same component. The important thing is
that provisions are made for incorporating electronic source records into some form of
recordkeeping system.
If an agency sends all its electronic records to a records management application
(RMA), NARA believes there is no need to build disposition functionality into its
word processing application or into a web tool that can search and retrieve
documents from the RMA.
Appropriate disposition must be applied to all records. In an ideal recordkeeping
situation, all documents created electronically would be under the control of the RMA. If
there are documents outside the control of the RMA, however, their disposition must also
16
be addressed. They cannot simply be assumed to be nonrecords. It is not clear how one
would build "disposition functionality into its word processing application." One would
most likely have to instead build in disposition functionality into the agency's overall
recordkeeping system, in order to address any documents that are not within the control
of the RMA.
What do we do about systems used to produce electronic records that are only
maintained in hard copy?
Disposition could still be considered at the design stage. An agency could identify that
the electronic information system will not carry out the disposition actions itself and
instead make provisions for doing so through other means.
3A3. How should “enhancements to existing systems” be defined or qualified to
indicate that new or different records are being created? NARA has a general
policy that agencies must reschedule their records when an agency program is
reorganized or otherwise changed in a way that results in the creation of new or
different records (see 36 CFR 1228.26(a)(2)).
NARA’s policy appears to address this issue adequately.
3A4. What activities does the term “produce” cover? Is there a clearer way to
state these activities?
“Produce” includes creation, revision, annotation and amendment. “Produce” seems to
be the simplest way to cover this concept. We would also suggest adding the word
"receive" to Public Citizen's list.
Do you see any other issues that should be considered as we evaluate Public
Citizen proposal 3?
We agree with Public Citizen that is extremely important to address records scheduling
and appraisal as early in the system development life cycle as possible, and this should
hold for all electronic records rather than simply data files. We share NARA's concern
over the wording proposed by Public Citizen, however, and feel that the provisions of
both 1228.26(a)(2) and 1234.32(a) address this point sufficiently.
We would like to reiterate a definition provided by the Electronic Records Work Group:
Business needs. An agency's need to conduct its business, maintain a
record of its essential activities and decisions for its own use, support
oversight and audit of those activities, and permit appropriate public
access. Agencies have certain responsibilities under the Electronic
Freedom of Information Act Amendments (EFOIA) to make records
available in electronic format. Although NARA does not have the
statutory authority to mandate how agencies comply with EFOIA,
17
agencies should be aware that public access is one of several business
needs that they need to consider in scheduling their electronic source
records.
As Public Citizen argues, retention of records in electronic formats will often greatly
promote both the convenience and quality of public access to those records. Recent case
law provides a number of examples in which parties have been required to provide the
electronic version of records as part of the evidence discovery process. 26 While this does
not imply that all records must be retained in electronic formats for their entire life, it
does provide compelling support for the view that electronic source records can often
contain important elements of evidence that may be lost in printed copies.
NARA's mission is ensuring "ready access to essential evidence." This is a primary
objective of the archival profession as a whole. It is also a mission in which the federal
government, citizens, and other potential researchers all have an interest. EFOIA
represents an important set of business needs related to such access. Recent legislation
such as the Government Paperwork Elimination Act27 and Electronic Signatures in
Global and National Commerce Act28 has also introduced a number of requirements to
manage and provide access to government records in electronic formats. Of course,
agencies must continually strike a balance among numerous business needs. This
balance is recognized in the FOIA legislation, which states that each "agency shall make
reasonable efforts to maintain its records in forms and formats"29 that can be both
searched and reproduced electronically. We encourage NARA to continue expanding its
efforts to promote the management and preservation of electronic records within the
federal government.
1
Public Citizen Litigation Group, "Public Citizen Petition to the Archivist to Amend GRS 20 and
Electronic Records Regulations," (October 31, 2000). Available at
http://www.publiccitizen.org/litigation/briefs/ERecords/articles.cfm?ID=672
2
Task Force on the Artifact in Library Collections, "The Evidence in Hand: The Report of the Task Force
on the Artifact in Library Collections," (Washington, DC: Council on Library and Information Resources,
2001). Available at http://www.clir.org/pubs/abstract/pub103abst.html
3
Frank Evans, Donald Harrison and Edwin Thompson, "A Basic Glossary for Archivists, Manuscript
Curators, and Records Managers," American Archivist 37 (1974): 424.
4
Committee on Intrinsic Value, "Intrinsic Value in Archival Material," Staff Information Paper Number 21
(Washington, DC: National Archives and Records Service, 1982).
5
James M. O'Toole, "On the Idea of Uniqueness," American Archivist 57 (1994): 632-58.
6
36 CFR 1234.30. Selection and maintenance of electronic records storage media.
7
Charles M. Dollar, Authentic Electronic Records: Strategies for Long-Term Access (Chicago, IL:
Cohasset Associates, 1999).
8
Kenneth Thibodeau, "To Be or Not to Be: Archives for Electronic Records," In Archival Management of
Electronic Records, edited by David Bearman (Pittsburgh, Pa: Archives and Museum Informatics, 1991),
1-13.
9
Abby Smith, ed. Authenticity in a Digital Environment (Washington, DC: Council on Library and
Information Resources, 2000).
10
Lewis J. Bellardo and Lynn Lady Bellardo, A Glossary for Archivists, Manuscript Curators, and Records
Managers, Archival Fundamentals Series (Chicago: Society of American Archivists, 1992), 19.
18
11
2 F. Supp. 2d; see also 184 F. 3d at 910-11
36 CFR 1228.270(d)(4)
13
36 CFR 1228.270(e)
14
David Holdsworth, and Derek M. Sergeant, A Blueprint for Representation Information in the OAIS
Model (2000). Available at http://gps0.leeds.ac.uk/~ecldh/cedars/nasa2000/nasa2000.html.
15
RLG-OCLC Digital Archive Attributes Working Group, "Attributes of Trusted Repositories for Digital
Research Resources: Meeting the Needs of Research Resources," (Mountain View, CA: Research Libraries
Group, 2001). Available at http://www.rlg.org/longterm/attributes01.pdf
16
44 USC 3303
17
36 CFR 1234.2, under "data file" definition.
18
One could argue that relational database tables are more "strictly prescribed" than electronic mail
messages, though this may not always be the case. Relational databases generally support large free-text
fields, and binary-large objects (BLOBs). They also often contain stored procedures, triggers and pointers
to external applications, all of which reach far beyond the bounds of the relational data model itself.
Though electronic mail messages are generally described as "semi-structured," their allowable data
elements are actually quite formally defined. In fact, both client and server electronic mail applications
often store and manage messages within relational databases. As both database and electronic mail
applications increasingly provide native XML support, the structured/semi-structured distinction between
the two blurs even further. The important distinction for purposes of recordkeeping is the nature of
creation and use. NARA quite rightly provides a handful of required data elements for electronic mail
messages, since the way in which they are used tends to both require and yield those elements, regardless
of the record series under which the messages are categorized.
19
Bellardo and Bellardo, 8.
20
For a description of the concepts of form and function (within the context of description and access,
rather than records management), see David A. Bearman and Richard H. Lytle, "The Power of the Principle
of Provenance," Archivaria 21 (1985): 14-27.
21
For some simple formats, this will be a "body" element that is not further differentiated.
22
National Archives and Records Administration, "Introduction to the General Records Schedules,"
(December 1998). Available at http://ardor.nara.gov/grs/grsintro.htm
23
There is a great deal of digital data associated with a record that is hidden from the view of most users
but can be revealed through the use of forensic tools. These include scratch files, "undo" text stored within
files, various caches, history files, temporary files, and even data that has been deleted but can still be read
off of the physical medium. While the proper destruction of such data can be a vital security concern in
some environments, their ongoing retention is rarely a recordkeeping requirement, since they serve only to
facilitate efficient operations (error recovery, crash protection, etc.) rather than contributing directly to the
content, structure or context of a record as evidence of a given function, activity, transaction or event.
24
For an analysis of many properties of concern in the preservation of spreadsheets, see Gregory W.
Lawrence, William R. Kehoe, Oya Y. Rieger, William H. Walters, and Anne R. Kenney, "Risk
Management of Digital Information: A File Format Investigation," (Washington, DC: Coalition on Library
and Information Resources, 2000). Available at http://www.clir.org/pubs/abstract/pub93abst.html
25 John Jessen, "The Continuing Impact of Electronic Discovery," Paper presented at the National
Conference on Managing Electronic Records (Chicago, IL, September 24-26 2001).
26
See e.g. Greyhound Computer Corp., Inc v. IBM 3 Computer L. Serv. Rep. 138, 139 (D. Minn. 1971);
Adams v. Dan River Mill, Inc. 54 F.R.D. 220 (W.D. Va. 1972); Minnesota v. Philip Morris Inc., No. CI-948565 (Dist. Ct. Minn.); National Union Electric Corp. v. Matsushita Electric Industrial Co., 494 F. Supp.
1257 (E.D. 1980); Williams v. Owens-Illinois, Inc., 665 F.2d 918 (C.A. 9, 1982); In re Air Crash Disaster,
130 F.R.D. 634 (E.D. Mich. 1989); State of New York and UDC-Love Canal Inc. v. Hooker Chemicals and
Plastics Corp, Order, CIV-79-990 (W.D.N.Y. Nov. 30, 1989).
27
Public Law 105-277, Division C, Title XVII
28
Public Law 106-229
29
5 USC 552(a)(3)
12